System and method for splitting a replicated data partition

摘要

A system that implements a data storage service may store data on behalf of storage service clients. The system may maintain data in multiple replicas of partitions that are stored on respective computing nodes in the system. The system may split a data partition into two new partitions, and may split the replica group that stored the original partitions into two new replica groups, each storing one of the new partitions. To split the replica group, the master replica may propagate membership changes to the other members of the replica group for adding members to the original replica group and for splitting the expanded replica group into two new replica groups. Subsequent to the split, replicas may attempt to become the master for the original replica group or for a new replica group. If an attempt to become master replica for the original replica group succeeds, the split may fail.

主权项

1. A system, comprising:
a plurality of computing nodes, each comprising at least one processor and memory, wherein the plurality of computing nodes is configured to implement a data storage service; wherein the data storage service maintains data on behalf of one or more storage service clients, wherein the data is partitioned into one or more partitions, and wherein said maintaining the data comprises maintaining two or more replicas of the data stored in a given partition on respective computing nodes in the system, wherein the two or more replicas make up a replica group, wherein at most one of the replicas in the replica group can act as a master replica for the replica group at a time, wherein replicas in the replica group that are not acting as the master replica act as slave replicas in the replica group, and wherein one of the replicas is configured to act as master replica for the replica group; wherein the master replica is configured to:
propagate metadata indicating one or more membership changes for the replica group to the other replicas of the replica group, wherein each of the one or more membership changes comprises adding one or more replicas to the replica group; andpropagate metadata indicating a subsequent membership change for the replica group to the other replicas of the replica group, wherein the subsequent membership change comprises splitting the original replica group into two new replica groups, and wherein the original replica group comprises the members of the replica group immediately prior to the membership change that splits the replica group; and wherein, for at least one of the new replica groups, a replica included in the new replica group is configured to attempt to assume the role of master replica for the new replica group, wherein said attempting comprises:
gathering state information from at least some of the other replicas in the new replica group; anddetermining, dependent on the state information, whether at least a pre-determined number of other replicas in the new replica group support the attempt to assume the role of master replica for the new replica group.